AI tools, scored for your job
Get paid to train AI
🤝 You're in the Free & Open Source library — community-built tools you can use without paying. A different concept from our scored AI directory.
Home › Free & Open Source › Students › FunASR
✦ Free & open sourceStudents

FunASR

Industrial speech recognition toolkit for ASR, VAD, and emotion models.

FunASR Speech Recognition

Industrial speech recognition toolkit with offline, streaming, and edge deployment.

Free forever · open source · MIT
✓ Light setup✓ Browser-based✓ Real-time collaboration
7.9
ZEKAI
Powerful speech toolkit
Free & open source · Students
Ease of setup4
Usefulness9
Maintenance9.5
Community9.2
★ 20,524
GITHUB
Sep 2026
UPDATED
MIT
LICENSE
🛠 Don't want to self-host? We can set it up →
⚡ Quick answer

FunASR is an industrial-grade open-source speech recognition toolkit supporting training, inference, streaming ASR, VAD, punctuation, and speaker diarization. It suits developers and researchers who need high-performance local audio transcription.

Runs on
Cross-platform
Hosting
Self-host
Languages
Python
License
MIT

What it is

The toolkit provides models like SenseVoice and Paraformer for multilingual transcription, voice activity detection, emotion recognition, and audio-event classification. It supports Python integration, a CLI for agents, and GGUF binaries via llama.cpp for CPU and edge devices without Python.

Who it's for

It suits developers building localized speech applications or researchers needing precise Chinese and multilingual speech pipelines. It does not suit end users looking for a plug-and-desktop GUI dictation app without command-line usage.

Key features

OpenAI-Compatible Serving

Run local endpoints matching OpenAI audio transcription APIs for easy integration.

Streaming Real-Time ASR

Feed audio chunk by chunk to generate live transcriptions with custom chunk sizes.

Emotion Recognition

Analyze audio files using emotion2vec models for utterance-granularity emotion detection.

Edge GGUF Binaries

Run models as a single self-contained binary via llama.cpp without needing Python.

How to start

Up and running in 3 steps

1
Install Python dependencies including torch, torchaudio, and funasr via pip
2
Initialize the AutoModel class in Python with your chosen model, VAD, and device settings
3
Run transcription locally using the Python API, CLI commands, or by launching the local HTTP server
Real talk

Our honest verdict

What's great

  • Significantly lower Chinese CER compared to whisper.cpp in benchmark tests
  • Built-in FSMN-VAD and speaker diarization pipelines
  • Flexible deployment options from Python SDK to CLI and Docker containers
  • Supports hardware acceleration via CUDA and Vulkan backends

Where it falls short

  • Requires manual setup of Python environments, PyTorch, and specific CUDA drivers
  • Running GPU-accelerated servers demands compatible NVIDIA hardware and vLLM configuration
  • Documentation and guides are split across multiple markdown files and scripts
Cost

Pricing & license

Free under its MIT licence, with no paid hosted options mentioned in the repository. The only cost is running it yourself on your own hardware.

Paid work

Open remote AI-training roles

880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.

See all open roles →
Compare

Alternatives

🎁 Free & open-source alternatives
Get more

Use it, or have us set it up

📄 Free PDF: FunASR for Students
A plain-language guide to getting your first result. Sent to your inbox.
🔔 64 people want a full video tutorial — add your vote
🛠 Don't want to set it up yourself?
We'll install, host & configure FunASR for you — and train your team.
Get it done for you ↗
Questions

FAQ

Can I run FunASR without Python installed?

Yes, you can use the prebuilt llama.cpp GGUF binaries to run models like SenseVoice as a single self-contained binary on CPU or edge devices.

Does FunASR support real-time streaming audio?

Yes, you can feed audio chunk by chunk into the streaming model configuration to output real-time text results.

Radar

Latest AI news

✉️ The weekly free toolkit
The best new free & open-source tools for Students — one short email a week, in your language.
Free · 1 email/week · unsubscribe anytime
Get paid to train AI →

See Zekai first in Google